Five Findings

1. 1.Using AI and understanding AI are different skills.

   One of the clearest gaps I encountered was between knowing how to use an AI tool and knowing how to evaluate what it produces.

   Library patrons increasingly encounter AI-generated summaries, search results, images, citations, and other information without necessarily knowing how to determine whether those outputs are reliable.

   This makes traditional information-literacy skills newly important: checking sources, corroborating claims, tracing provenance, identifying hallucinations, and distinguishing a plausible answer from a verified one.

   “The muscles around doing horizontal reading—looking something up, checking it in another browser tab, seeing if it is corroborated elsewhere—aren’t always there.”

2. 2.Start with what people are actually trying to do.

   People appeared more interested in learning about AI when it was connected to an existing need: finding a job, writing a résumé, creating content, improving workplace productivity, evaluating news, or staying safe online.

   That suggests a different approach to AI education. Instead of beginning with “What is a large language model?”, a public resource might begin with questions such as:

   • How can I use AI during a job search?
   • How do I verify an AI-generated answer?
   • What information shouldn’t I give a chatbot?
   • How can I determine whether an image or article is trustworthy?

   Technical concepts can then be introduced when they help answer those questions.

3. 3.Information alone doesn’t replace human guidance.

   Another recurring finding was the importance of trusted intermediaries.

   People don’t always want another webpage explaining AI. They may want to show someone an unfamiliar output, ask a follow-up question, or talk through uncertainty.

   “People are looking for something where they can ask a question, have a follow-up, or talk to somebody.”

   Libraries are particularly interesting in this context because they combine access to information with workshops, one-on-one technical assistance, workforce services, and information-literacy education. Rather than replacing that human support, an AI-literacy resource could provide a shared foundation for conversations among librarians, educators, community organizations, and the public.

4. 4.AI literacy should increase agency—not simply AI adoption.

   There was no single “public attitude” toward AI in my research. Librarians encounter enthusiastic adopters, skeptics, complete beginners, technical users, and many people somewhere in between.

   Meanwhile, AI is increasingly embedded in services people already use. That makes choice an important component of AI literacy.

   “I don’t think anyone should feel pressure to use any technology, the same way that we shouldn’t be telling anyone not to use something.”

   The goal of AI education, then, doesn’t have to be convincing people to embrace or reject AI. It can instead give people enough understanding to decide when to use it, when to question it, and when not to use it at all.

5. 5.AI literacy begins with digital literacy.

   AI education cannot assume that everyone starts with the same technical knowledge.

   Some library patrons are already asking sophisticated questions about model differences, data practices, environmental effects, and ethics. Others have heard terms such as “AI” and “ChatGPT” without understanding what they describe.

   Privacy, security, misinformation, scams, and basic information verification therefore aren’t side topics. They are prerequisites.

   “Foundational technical literacy and understanding of privacy and security have to be baked into all of this, too.”